What challenges did The Vanguard Group, Inc. face before AI implementation?

Analysts needed faster, more direct access to financial data for decision-making. The existing workflow required SQL expertise and data team support, with typical requests taking several days to fulfill. AI needed reliable enterprise data foundations, semantic context, and metadata management to generate accurate business-relevant insights.

What was the impact of the AI solution?

Reduced time-to-insight from days to minutes. Enabled business users to access data independently without SQL knowledge. Achieved high accuracy in AI-generated SQL queries. Decreased data team workload for routine analytical requests. Established a reusable framework adopted across multiple Vanguard business units.

Rank#923of 3,449

Source

ExploringEvidence: Low40/100

Vanguard Virtual Analyst: conversational financial-data access with Amazon Bedrock

Use case typeDecision supportUpdated Apr 29, 2026

Vanguard built a Virtual Analyst for analysts and business stakeholders to query complex financial datasets through natural language. The implementation relies on AI-ready data foundations, including a metadata catalog, semantic layer, ground-truth question-to-SQL examples, automated data quality checks, and AWS services such as Amazon Bedrock, Amazon Bedrock Guardrails, Amazon ECS, Amazon S3, AWS Glue, and Amazon Redshift.

Organization: The Vanguard Group, Inc.
Industry: Finance
Location: United States
Published: April 2026

Reported outcomes

Strategic outcomes

Better decisions & insightReduced time-to-insightOther strategic outcomeEnabled self-service data accessCustomer experience & trustImproved SQL accuracyOther strategic outcomeReduced data team workloadScale & capacityReusable across business units

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: The Vanguard Group, Inc.
Provider: AWS
Maturity: Exploring
Linked source: AWS Machine Learning Blog

Implemented a unified metadata catalog, semantic layer, ground-truth exemplars, automated data quality checks, change control, and continuous evaluation

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Decision support
2Data governance
3Conversational assistants

Analysts needed faster, more direct access to financial data for decision-making.
The existing workflow required SQL expertise and data team support, with typical requests taking several days to fulfill.
AI needed reliable enterprise data foundations, semantic context, and metadata management to generate accurate business-relevant insights.

Built an AI-ready data architecture and operating model for the Virtual Analyst.
Implemented a unified metadata catalog, semantic layer, ground-truth exemplars, automated data quality checks, change control, and continuous evaluation.
Used Amazon Bedrock for foundation models, Bedrock Guardrails for input/output protection, Amazon ECS for compute, Amazon S3 for persistence, AWS Glue for cataloging and ETL, and Amazon Redshift for centralized data warehousing.

Technologies

Amazon Bedrock Amazon Bedrock Guardrails Amazon ECS Amazon S3 AWS Glue Amazon Redshift

Reduced time-to-insight from days to minutes.
Enabled business users to access data independently without SQL knowledge.
Achieved high accuracy in AI-generated SQL queries.
Decreased data team workload for routine analytical requests.
Established a reusable framework adopted across multiple Vanguard business units.

Architecture

Vanguard built an AI-ready data architecture and operating model with a unified metadata catalog, semantic layer, ground-truth question-to-SQL exemplars, automated data quality checks, change control, and continuous evaluation; the solution uses Amazon Bedrock, Amazon Bedrock Guardrails, Amazon ECS, Amazon S3, AWS Glue, and Amazon Redshift.

Sources & evidence1

Evidence: Low40/100Evidence strength